Intel Not Taking Any Chances with Core i7

[Veri]

Intel, Intel, give me your answer do,
Going crazy, can't divide 3 by 2.
My answers I can't see 'em,
They're stuck in my Pent-i-um,
So you'd look great,
If you would make,
A functional FPU!

Or something like that, original author forgotten.

From an engineering PoV, it's maybe more impressive that relatively unflawed chips got out before the methodology-changing Pentium debacle than is the amount of make-110%-sure testing that goes into chips today. Formal methods is first year comp sci (specific invariants have only been researched in the last decade or so, but the opportunity has been there much longer), so Intel's almost passable original choice of divine insight is to be commended ;-).

 

[hal1984]

Apple could mount mobile quad core penryns (45 W TDP) in the iMacs and MacBook Pro to maintain a high performance more or less close to Core i7.

 

[MrZebra]

Apple could mount mobile quad core penryns (45 W TDP) in the iMacs and MacBook Pro to maintain a high performance more or less close to Core i7.

and then they would set your house on fire

 

[adrian.oconnor]

This is exactly my point! Why do we label "bugs" in cars as mediocre design and why an "unavoidable fact" in software engineering?

I missed what you were trying to say, sorry. In that case, good point

I am not convinced about the "price" and "time" argument. I believe it is a mind shift.

Well, I can kind of agree with that up to a point, but I don't see how it can happen in the real world.

The reason cars are so reliable is, I believe, due to several things: very strict homologation, very long-term gradual changes to existing designs and wide spread standardisation of parts.

While it would be nice to apply these things to software, to try and do so would effectively slam the brakes on the speed of innovation. Consumers seem willing to take the downsides of rapidly developed products because it means they can have new technologies sooner (and cheaper).

As it happens, I write business software for a living. I know first hand that many companies would prefer to take a buggy product early than wait for a stable final version because they can start benefiting from it right away. I think this has been documented elsewhere as a reason that buggy software gets shipped -- it's because it still has a net positive value. Of course, I personally never ship buggy software

In the 1980's US and Europe manufactured cars were full of "bugs". We thought those "bugs" were an "unavoidable fact" of car manufacturing.
The Japanese brands changed our way of thinking. Their concept of quality changed the entire industry. This did not lead to an explosion of car prices, this did not lead to a slowing in the new-model-cycle.

Audi, BMW and Mercedes made some wonderfully robust cars in the 1980s and in many ways were the model that Japanese companies tried to emulate. I'm not sure how the Japanese actually managed to do it in such a major way, but I've read several times that it had a lot to do with treating their workers as the skilled and valuable people that they were, and not as cogs.

I am British, and all of the companies that I have worked for have treated me very well. However, I understand that many programmers in America are put in cubicles and treated very badly - squeezed for every last second of time that they can be forced to work. That would seriously affect my ability to create high quality software, and maybe that goes someway to explaining things?

(I don't think Intel stick their chip designers in cubicles, so your i7 MacPro should still be good )

 

[AidenShaw]

I was tempted to ask how much for your P5E3 but it's X38 and DDR3.

Yes, it's last November's technology ...

By the way, the pricing looks good:

Now, where's that Apple mini-tower with this chip????

 

[diamond.g]

Yes, it's last November's technology ...

By the way, the pricing looks good:

Now, where's that Apple mini-tower with this chip????

It is right behind the Powerbook G5....


Seriously it looks like the cost of the higher end chip would keep Apple from putting it in a sub 2500 dollar machine.

 

[840quadra]

Let's just hope they're a lot more efficient than the PowerPC chips, though, right?

The G5 yes.

Back when the G3, and G4 were new, they were the chips to beat.

 

[Joe The Dragon]

Yes, it's last November's technology ...

By the way, the pricing looks good:

Now, where's that Apple mini-tower with this chip????

Right now you are looking at a $300 cpu + a $300 MB for core i7 + 3 ddr 3 ram dimms with low voltages.

Where are the lower end core i7 chip sets?

 

[AidenShaw]

Right now you are looking at a $300 cpu + a $300 MB for core i7 + 3 ddr 3 ram dimms with low voltages.

Where are the lower end core i7 chip sets?

And you are looking at list prices. Gateway's selling a system that will stomp an $1800 Imac for only $1250...

Newegg is selling X58 mobos for $221.

Intel usually releases higher end parts first, then broadens the line as the prices come down.

Note also that Newegg is selling the 2.8GHz quad-core Xeons for $730 each, yet a Mac Pro with two of them is $2800. That should be a good gauge of what Apple's discounts used to be. ("used to be", because Intel might not be so generous in the future after all the noise about the Nvidia chipsets in the MacBooks )

 

[Firefly2002]

When can we expect Nehalem in a Mac?

Nehalem is probably one of the biggest upgrades a Mac will ever see.

One of the dumbest things I've ever heard. 68040? The PowerPC? The 604? G3? G5?

We won't have it for a bit.

Anyway, I don't see the point of this. The issue with the FPU in the original Pentium wasn't an enormous deal and not worth 500 mil.... most bugs are almost impossible to even stumble upon.

The G5 yes.

Back when the G3, and G4 were new, they were the chips to beat.

The G5 was too, at least when it first came out (quickly ceased to be true) especially clock for clock, though AMD was pretty close... Intel of course wasn't even in the ballpark.

Don't forget the rest of the PowerPCs... they were all also the chips to beat (and were unbeatable)

 

[reinman78]

This is why I did not buy the mbp, waitng for core i7.





IMAC 2.8 ghz intel core 2 extreme
AEBS
IPHONE 3g

 

[macsmurf]

Formal methods is first year comp sci (specific invariants have only been researched in the last decade or so, but the opportunity has been there much longer), so Intel's almost passable original choice of divine insight is to be commended ;-).

As far as I know, proving formally that code works the way it's supposed to is inherently difficult and extremely time consuming for anything more than the Mickey Mouse examples that one might encounter in a first year comp sci course, which is why it hasn't been widely adopted in the industry.

Hardware may be a different ball of wax altogether, but in any case it certainly wouldn't be first-year comp sci stuff.

 

[hal1984]

and then they would set your house on fire

I don't understand.

 

[kabunaru]

One of the dumbest things I've ever heard. 68040?

Why so nitpicky?
I meant one of the biggest upgrades the Intel Macs will ever see.

 

[mjteix]

Right now you are looking at a $300 cpu + a $300 MB for core i7 + 3 ddr 3 ram dimms with low voltages.

Where are the lower end core i7 chip sets?

There will be no low-end X58 chipset (if that's want you meant). There's only one X58 Express chipset and it has 36 PCIe lanes.

Also did you get the news about psystar?

 

[iMacmatician]

I think you lack an understanding of software engineering. To just throw down a blanket claim that mediocre design, development or testing is responsible for bugs is just plain silly.

I've heard that it's mathematically proven that software cannot be bug-free, no matter what.

Apple could mount mobile quad core penryns (45 W TDP) in the iMacs and MacBook Pro to maintain a high performance more or less close to Core i7.

Probably not the MacBook Pro, but the iMac definitely. The only thing holding Apple back is the decision to put lower-clocked quad-cores with higher-clocked dual-cores in the same product line.

 

[mjteix]

Probably not the MacBook Pro, but the iMac definitely. The only thing holding Apple back is the decision to put lower-clocked quad-cores with higher-clocked dual-cores in the same product line.

Unless, they offer the following:
$1199 20" iMac dual-core 2.40GHz (LED-BL display, 2GB RAM,...)
$1499 20" iMac quad-core 2.00GHz, low-end GPU
$1799 24" iMac quad-core 2.00GHz, mid-range GPU
$2199 24" iMac quad-core 2.26GHz, mid-range GPU
$2499 24" iMac quad-core 2.53GHz, high-end GPU (25th anniv. Mac)

 

[uricmu]

Well, the news is pretty sparse on details but mention formal methods, which is a completely different approach from 'random' testing. I was not aware that it was a big thing in hardware development still. If someone has some references I would love to see them.

In the late 90s they were doing both since you can obviously get more coverage with random testing than with formal methods. Formal methods were useful or testing certain behaviors and protocols or stateful components such as memory controllers. I would venture a guess that they are still doing the same with Nehalem and that the NY piece is more of a PR thing than anything new.

 

[xbjllb]

Intel, Intel, give me your answer do,
Going crazy, can't divide 3 by 2.
My answers I can't see 'em,
They're stuck in my Pent-i-um,
So you'd look great,
If you would make,
A functional FPU!

Or something like that, original author forgotten.

Anyone besides me ever notice that "Bicycle Built For Two" and "The Sidewalks Of New York" were practically the same damn song?

 

[macsmurf]

I've heard that it's mathematically proven that software cannot be bug-free, no matter what.

It hasn't. In fact, it IS possible to prove that a certain program is bug-free, and for very small programs it is almost trivial.

However, there are several problems when one tries to do it in the real world. The US military tried to do formal verification of programs with regards to security in the 1980s. It turned out that the first mathematical models of security were not sophisticated enough to prove anything interesting and the more complex the model became, the more complex the proof process became to the point that only a few highly trained specialist were able to do it, and even they could only manage to verify a couple of lines of code a day.